Search Results

You are looking at 61 - 70 of 851 items for :

  • Deep learning x
  • All content x
Clear All
Donat Perler and Oliver Marchand

1. Introduction This study investigates adaptive boosting, a relatively new classification method introduced by Freund and Schapire (1997) , to numerical weather prediction (NWP) output postprocessing. As a case study, we show how this machine learning method can be used to detect thunderstorms in NWP forecast output fields. NWP uses the basic physical equations of the atmosphere for simulation. Predicting thunderstorms with NWP models is inherently difficult. This is because of the rather

Full access
Tom H. Durrant, Diana J. M. Greenslade, Ian Simmonds, and Frank Woodcock

spatially homogeneous corrections, to corrections that vary both in space and time. In an effort to eliminate the need to manually monitor and update the applied corrections, an automatic, self-learning correction method is proposed, applicable to operational forecast winds. This work was performed with the intention of developing an operational system applicable to the Bureau’s forecasting environment. The work was carried out within the context of replacing the Bureau’s operational wave model

Full access
Scott E. Kalafatis, Jasmine Neosh, Julie C. Libarkin, Kyle Powys Whyte, and Chris Caldwell

composed of their own unique blend of reflective activities to continually monitor and enhance their work. This finding highlights that even though there were suggested practices offered and documented in this study, effective learning by doing is a deeply personalized process where participants’ past history, current experiences, and future goals come into contact with one another in the service of personal and professional growth. Suggested practices have reflective value as objects or ideals that

Full access
Bruno Buongiorno Nardelli and Rosalia Santoleri

SST measurements to define bins on which different GEM fields can be estimated ( Mitchell et al. 2004 ). The number of DOFs absorbed by the GEM techniques is 1 (given by the tension applied to the curve) plus the number of nodes used in the smoothing. The latter is generally chosen to be the maximum between a fraction of the total number of profiles in the learning dataset and a minimum of 3. As a consequence, the DOFs absorbed by a basic GEM are generally higher than 4. If different bins are

Full access
Andrew E. Mercer, Alexandria D. Grimes, and Kimberly M. Wood

learning method could benefit from considering multiple processes simultaneously as opposed to the best-separating field presented here (a task outside of the scope of this project). Typically, the best-separating GFSA fields were consistent among multiple RI definitions, revealing humidity across a deep vertical layer, low-level instability, and midlevel vorticity as potentially important separating fields for RI/non-RI environments. Ultimately, the improved separability offered by the presented

Restricted access
Paul J. Croft and Juyoung Ha

happened. This was truly one of the examples of which students took the leadership in designing and developing the course and became professionally competent through immersion. It also marked their increasing comprehension of what it means to be a scientist rather than simply doing science. This unique hands-on educational experience prepared and motivated students to be leaders of the project and their learning because they were now committed to strive toward deeper understanding and completion of the

Full access
Dawn Kopacz, Lindsay C. Maudlin, Wendilyn J. Flynn, Zachary J. Handlos, Adam Hirsch, and Swarndeep Gill

for educators. Despite the interest in ASER activities, many of the instructor respondents perceive at least a moderate level of risk to their promotion and/or tenure potential if they decide to engage in ASER activities through collaborations on teaching and learning research projects and presentations at discipline-based conferences. The lack of interest in becoming more deeply involved in the ASER community may be a result of a perceived lack of value placed on ASER by their institutions. This

Full access
S. Caires and A. Sterl

necessity of having a learning dataset (TOPEX measurements) in all the periods with different error characteristics and the prerequisite knowledge of the periods with a given type of error characteristics. There were some caveats in the ERA-40 data that remain in the C-ERA-40 dataset. The ERA-40 model does not account for shallow-water effects and therefore the data are only valid in deep-water regions. Due to resolution, tropical cyclones are not resolved by the ERA-40 system. Therefore, the data

Full access
Wiebke Schubotz, Daniel Klocke, Ulrich Löhnert, Andreas Macke, Bjorn Stevens, and Allison Wing

observational data are used in synergy with modeling efforts. Another hot topic was machine learning, as it came up in several presentations at UCP2019. These showed the potential to learn parameterizations by training deep neural networks and the use of artificial intelligence to probe inputs and outputs of existing parameterizations. They also highlighted how computationally expensive parts of a model can be replaced by faster algorithms. Another example of how machine learning approaches can be leveraged

Free access
Robin L. Tanamachi, Daniel T. Dawson II, and Loran Carleton Parker

learning of atmospheric science, we created an elective “severe storms field work” course within Purdue EAPS. Learning, in this context, is defined as, “the process whereby knowledge is created through the transformation of experience” ( Kolb 1984 ). The rationale for the creation of this course was well expressed by King (1993) : “When students are engaged in actively processing information by reconstructing that information in such new and personally meaningful ways, they are far more likely to

Free access